METHODS & RESPONSIBILITIES · SEPTEMBER 19, 2026
What the checking covers.
An automated report is an observation of particular pages in particular states. Its coverage record matters as much as its findings.
Start with a written scope.
Two options: the $199 fix-ready pack (documented findings with exact locations, written for your developer), or the $849 done-for-you fix sprint (we fix them in an unpublished copy of your theme — you approve every change before anything goes live, with one-click rollback). Final scope, pages, price and timing are agreed in writing before work begins.
A product URL is not every state of that product page. An opened cart drawer, selected variant, menu, modal, login flow, or checkout needs separate evaluation. The current service checks loaded pages without exercising those interactions.
Browser automation and axe-core.
Playwright opens the agreed URLs in Chromium. axe-core runs checks tagged WCAG 2 A/AA, WCAG 2.1 A/AA, and best practice. Those tags select automated rules; they do not mean every requirement in those standards has been tested.
One recorded state
The loaded page is examined at a recorded desktop viewport. Content behind an interaction, a login, or a challenge is not automatically covered.
A point in time
Results include the run date, tool versions and observed URLs. Later content, theme or app changes can produce different results.
Read the evidence in context.
- Coverage: requested and final URLs, page outcomes, errors and the state examined.
- Findings: rule IDs, axe impact labels, affected-node counts and example selectors, markup and tool explanations.
- Investigation: tool documentation and next checks to consider. These are not verified fixes.
- Incomplete checks: results for which the tool could not reach a conclusion, kept separate from violations.
For each rule on each page, the report shows up to three violation examples and up to three incomplete-check examples. The saved scan record retains the underlying returned results. The sample shows the current report structure on a controlled fixture; customer coverage must be agreed separately.
Counts are observations, not scores.
An instance is an affected node returned under one rule on one successfully checked page. A shared component can count again on another page, and an element can appear under more than one rule. Grouping by rule does not turn repeated instances into unique storefront problems.
Axe labels such as “critical” and “serious” are tool impact labels. Ordering by those labels is not a human assessment of business priority or legal risk.
A blocked or failed page is listed as unchecked. Its missing result is not zero findings. Incomplete tool checks remain unresolved. Even a completed check with no violations does not establish accessibility.
We check, then we fix.
The offer covers automated checking, documented findings, and done-for-you fixes in an unpublished copy of your theme. Your approval is required before anything goes live, and every change has one-click rollback. Manual keyboard and screen-reader testing, usability evaluation, rechecks beyond the agreed scope, ongoing monitoring and legal advice are outside the current offer.
Automated findings can point to possible barriers; they do not independently confirm how someone experiences a store. Qualified manual evaluation is needed to investigate those questions. This service does not provide legal compliance determinations or lawsuit protection.
Questions after the fixes.
Email Ari with the fix date, page URL and rule ID you want to discuss. Clarification arrangements are agreed in the written scope. Additional investigation and rechecking beyond the agreed scope are not implied, and no response time is guaranteed.